Knowledge representation systems and methods incorporating inference rules

• US 10,002,325 B2
• Filed: 01/06/2012
• Issued: 06/19/2018
• Est. Priority Date: 03/30/2005
• Status: Active Grant

• Current Assignee: Primal Fusion Inc.
• Original Assignee: Primal Fusion Inc.
• Inventors: Sweeney, Peter Joseph, Ilyas, Ihab Francis, Zhou, Wu, Oldford, Wayne
• Primary Examiner(s): Gurmu, Muluemebet
• Application Number:
  US13/345,644
• Publication Number:
  US 20120166373A1
• Time in Patent Office: 2,356 Days
• Field of Search: 707758
• CPC Class Codes: G06F 16/84 Mapping; Conversion
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

Techniques for analyzing and synthesizing complex knowledge representations (KRs) may utilize an atomic knowledge representation model including both an elemental data structure and knowledge processing rules stored as machine-readable data and/or programming instructions. One or more of the knowledge processing rules may be applied to analyze an input complex KR to deconstruct its complex concepts and/or concept relationships to elemental concepts and/or concept relationships to be included in the elemental data structure. One or more of the knowledge processing rules may be applied to synthesize an output complex KR from the stored elemental data structure in accordance with context information. Methods of populating an elemental data structure and methods of synthesizing complex KRs from the elemental data structure may rely on linguistic inference rules and/or elemental inference rules.

Claims

1. A method of modifying a computer-readable elemental data structure of a knowledge representation system, the method comprising:

applying, using at least one processor executing stored program instructions, one or more rules of analysis to deconstruct a knowledge representation into one or more elemental components;

adding, using at least one processor executing stored program instructions, data associated with the one or more elemental components to an elemental data structure, the elemental data structure storing data representing concepts and concept relationships;

inferring, using at least one processor executing stored program instructions, candidate data associated with the elemental data structure,

wherein the inferring comprises detecting language in reference data,

wherein the language corresponds to a predetermined linguistic pattern expressing a relationship between at least a first concept and a second concept in natural language;

modifying the elemental data structure to combine the candidate data and the data associated with the one or more elemental components,

wherein the modifying comprises adding, to the elemental data structure, the first concept, the second concept, and/or the relationship between the first and second concepts expressed by the linguistic pattern detected in the reference data, wherein the reference data is obtained from a source other than the knowledge representation

wherein the detected linguistic pattern expresses in natural language that the second concept subsumes the first concept;

wherein detecting the language corresponding to the predetermined linguistic pattern comprises detecting in the reference data a first label associated with the first concept, followed by a subsumptive expression, followed by a second label associated with the second concept, wherein the linguistic pattern including the subsumptive expression expresses in natural language that the second concept subsumes the first concept;

wherein the subsumptive expression comprises at least one of one or more predetermined words or one or more predetermined symbols;

wherein detecting the predetermined linguistic pattern in the reference data comprises detecting in the reference data the first label associated with the first concept and the second label associated with the second concept, wherein a proximity of the first label to the second label is within a threshold proximity;

wherein the one or more elemental components are encoded as one or more computer-readable data structures storing data associated with the one or more elemental components, and

wherein the reference data is encoded as one or more computer-readable data structures storing data associated with reference communication.

2. The method of claim 1, wherein the subsumptive expression comprises at least one of “

is a”

, “

is an”

, “

is a field of”

, or “

is a type of”

.

3. The method of claim 1, wherein the threshold proximity is at least one of a number of words, a number of sentences, or a number of a paragraphs.

4. The method of claim 1, wherein the detected linguistic pattern expresses in natural language that the second concept defines the first concept.

5. The method of claim 4, wherein detecting the language corresponding to the predetermined linguistic pattern comprises detecting in the reference data a first label associated with the first concept, followed by a definitional expression, followed by a second label associated with the second concept, wherein the linguistic pattern including the definitional expression expresses in natural language that the first concept is defined by the second concept.

6. The method of claim 1, further comprising inferring second candidate data associated with the elemental data structure, the inferring the second candidate data comprising:

identifying a first elemental concept in the elemental data structure, the first elemental concept being defined by one or more first characteristic concepts;

identifying a second elemental concept in the elemental data structure, the second elemental concept being defined by one or more second characteristic concepts; and

determining that each characteristic concept in the one or more second characteristic concepts is in the one or more first characteristic concepts or subsumes a characteristic concept in the one or more first characteristic concepts.

7. The method of claim 1, wherein:

the elemental data structure comprises the first concept and the second concept; and

modifying the elemental data structure to combine the candidate data and the data associated with the one or more elemental components comprises to the elemental data structure a subsumptive relationship between the first concept and the second concept.

8. The method of claim 1, wherein the candidate data indicates that the second concept does not subsume the first concept.

9. The method of claim 8, wherein:

the elemental data structure comprises an elemental concept relationship between the first concept and the second concept, the elemental concept relationship indicating that the second concept subsumes the first concept; and

modifying the elemental data structure to combine the candidate data and the data associated with the one or more elemental components comprises one of removing the elemental concept relationship from the elemental data structure or reducing a probability associated with the elemental concept relationship in the elemental data structure.

10. A method of modifying a computer-readable elemental data structure of a knowledge representation system, the method comprising:

applying, using at least one processor executing stored program instructions, one or more rules of analysis to deconstruct a knowledge representation into one or more elemental components;

adding, using at least one processor executing stored program instructions, data associated with the one or more elemental components to an elemental data structure, the elemental data structure storing data representing concepts and concept relationships;

inferring, using at least one processor executing stored instructions, a candidate probability that an elemental concept relationship exists between a first concept and a second concept in the elemental data structure,

wherein the inferring comprises applying one or more elemental inference rules to the elemental data structure to compute a probability less than 100% that the elemental concept relationship exists, wherein the one or more elemental inference rules are applied to the elemental data structure in response to obtaining data indicating that a first label associated with the first concept and a second label associated with the second concept appear in reference data, wherein a proximity of the first label to the second label is within a threshold proximity; and

modifying the elemental data structure to combine the candidate probability and the data associated with the one or more elemental components, wherein the modifying comprises adding to the elemental data structure data representing the computed probability in association with the elemental concept relationship;

wherein the elemental concept relationship indicates that the second concept subsumes the first concept;

wherein applying the one or more elemental inference rules to the elemental data structure comprises;

identifying the first concept in the elemental data structure, the first concept being defined by one or more first characteristic concepts;

identifying the second concept in the elemental data structure, the second concept being defined by one or more second characteristic concepts; and

calculating, as the candidate probability, a probability that each characteristic concept in the one or more second characteristic concepts is in the one or more first characteristic concepts or subsumes a characteristic concept in the one or more first characteristic concepts;

wherein the one or more elemental components are encoded as one or more computer-readable data structures storing data associated with the one or more elemental components.

11. The method of claim 10, wherein the elemental concept relationship indicates that the second concept defines the first concept.

12. The method of claim 10, wherein the one or more elemental inference rules are applied to the elemental data structure in response to receiving context information from a user, the context information being associated with at least one of the first concept or the second concept.

13. The method of claim 10, wherein the one or more elemental inference rules are applied to the elemental data structure in response to obtaining data indicating that a label associated with at least one of the first concept or the second concept appears in reference data at a rate that exceeds a threshold rate.

14. The method of claim 10, wherein:

the elemental data structure comprises the elemental concept relationship and a previous probability associated with the elemental concept relationship; and

modifying the elemental data structure to combine the candidate probability and the data associated with the one or more elemental components comprises replacing the previous probability associated with the elemental concept relationship with one of the computed probability, an average of the computed probability and the previous probability associated with the elemental concept relationship, or a function of the computed probability and the previous probability associated with the elemental concept relationship.

15. A knowledge representation apparatus for modifying a computer-readable elemental data structure, the apparatus comprising:

one or more processors; and

a memory unit configured to store instructions which, when executed by the one or more processors, cause the one or more processors to perform a method comprising;

applying one or more rules of analysis to deconstruct a knowledge representation into one or more elemental components,

adding data associated with the one or more elemental components to an elemental data structure, the elemental data structure storing data representing concepts and concept relationships,

inferring candidate data associated with the elemental data structure,

wherein the inferring comprises detecting language in reference data,

wherein the language corresponds to a predetermined linguistic pattern expressing a relationship between at least a first concept and a second concept in natural language;

modifying the elemental data structure to combine the candidate data and the data associated with the one or more elemental components,

wherein the modifying comprises adding, to the elemental data structure, the first concept, the second concept, and/or the relationship between the first and second concepts expressed by the linguistic pattern detected in the reference data, wherein the reference data is obtained from a source other than the knowledge representation;

wherein the detected linguistic pattern expresses in natural language that the second concept subsumes the first concept;

wherein detecting the language corresponding to the predetermined linguistic pattern comprises detecting in the reference data a first label associated with the first concept, followed by a subsumptive expression, followed by a second label associated with the second concept, wherein the linguistic pattern including the subsumptive expression expresses in natural language that the second concept subsumes the first concept,

wherein detecting the predetermined linguistic pattern in the reference data comprises detecting in the reference data the first label associated with the first concept and the second label associated with the second concept, wherein a proximity of the first label to the second label is within a threshold proximity;

wherein the subsumptive expression comprises at least one of one or more predetermined words or one or more predetermined symbols.